Macro-based simulated floating-point programming environment for instrumentation applications

The need for high-accuracy arithmetic and wide dynamic range is obvious in several fields of digital instrumentation, e.g., adaptive control/filtering, parametric system identification, and arbitrary function generation. However, most of the existing low-cost signal processors have solely fixed-point capabilities. On the other hand, the latest true floating-point processors represent still too high expense in many embedded industrial applications. To solve the high-precision demand and the low-cost requirement simultaneously, we propose a simulated floating-point programming environment that can be utilized in applications where the primary signal band-width is no more than 5-10 kHz. The high-precision floating-point operations can be used together with the fast fixed-point instructions to provide the design engineer a convenient flexibility to use the fixed-point arithmetic in speed-critical functions, like high-order frequency selective prefiltering, and the simulated floating-point arithmetic in such special cases when either high accuracy or wide dynamic range is required.